Kamae: Bridging Spark and Keras for Seamless ML Preprocessing

TL;DR

Kamae is an open-source library that converts PySpark preprocessing pipelines into Keras models, ensuring consistent feature processing in production recommender systems and reducing engineering complexity.

Contribution

This work introduces Kamae, a novel tool that seamlessly translates Spark preprocessing steps into Keras layers, bridging offline and online ML environments.

Findings

Enables consistent preprocessing across training and inference.

Reduces engineering effort and risk of dataset shift.

Demonstrated on real-world datasets and pipelines.

Abstract

In production recommender systems, feature preprocessing must be faithfully replicated across training and inference environments. This often requires duplicating logic between offline and online environments, increasing engineering effort and introducing risks of dataset shift. We present Kamae, an open-source Python library that bridges this gap by translating PySpark preprocessing pipelines into equivalent Keras models. Kamae provides a suite of configurable Spark transformers and estimators, each mapped to a corresponding Keras layer, enabling consistent, end-to-end preprocessing across the ML lifecycle. Framework's utility is illustrated on real-world use cases, including MovieLens dataset and Expedia's Learning-to-Rank pipelines. The code is available at https://github.com/ExpediaGroup/kamae.